What causes OrcJavaInvocationException?

Java heap memory exhaustion ('java.lang.OutOfMemory', 'Java heap space') on the integration runtime — an older runtime version. Self-hosted Integration Runtime (SHIR) running out of resources while processing ORC files. Missing DB2 package for the user (SQLSTATE=51002 / SQLCODE=-805 in the error message). Concurrent copy runs on the same SHIR node exceeding available memory

How do I fix OrcJavaInvocationException?

If the error contains 'SQLSTATE=51002 SQLCODE=-805': bind the missing DB2 package for the user account running the pipeline.. If the error contains 'java.lang.OutOfMemory' or 'Java heap space': upgrade the Self-hosted Integration Runtime to the latest version.. After upgrading the SHIR, limit the number of concurrent copy runs on that runtime node to reduce memory pressure.. If using Azure IR, the Java memory is managed automatically — a retry after a few minutes may resolve transient OOM conditions.

Medium severityfabricMicrosoft Fabric →

Microsoft Fabric Error:
OrcJavaInvocationException

What does this error mean?

A Fabric Data Factory copy activity using the ORC format connector failed because of a Java invocation error. Common causes include Java heap memory exhaustion, an outdated self-hosted integration runtime version, or a missing DB2 package (SQLSTATE=51002).

Common causes

1Java heap memory exhaustion ('java.lang.OutOfMemory', 'Java heap space') on the integration runtime — an older runtime version
2Self-hosted Integration Runtime (SHIR) running out of resources while processing ORC files
3Missing DB2 package for the user (SQLSTATE=51002 / SQLCODE=-805 in the error message)
4Concurrent copy runs on the same SHIR node exceeding available memory

How to fix it

1If the error contains 'SQLSTATE=51002 SQLCODE=-805': bind the missing DB2 package for the user account running the pipeline.
2If the error contains 'java.lang.OutOfMemory' or 'Java heap space': upgrade the Self-hosted Integration Runtime to the latest version.
3After upgrading the SHIR, limit the number of concurrent copy runs on that runtime node to reduce memory pressure.
4If using Azure IR, the Java memory is managed automatically — a retry after a few minutes may resolve transient OOM conditions.

Beyond the docs

Common practitioner solutions not covered in the official documentation.

1For Self-hosted IR: check available memory on the host machine and consider scaling up or distributing the load across multiple SHIR nodes

Frequently asked questions

Does this error affect Power BI reports in the same workspace?

Depends on the error type. Semantic model failures affect report freshness directly. Dataflow or pipeline failures may cascade into downstream semantic model failures.

How is debugging Fabric errors different from classic Power BI errors?

Fabric errors often require checking capacity utilization alongside the item-level error. The Fabric admin portal shows capacity pressure that Power BI Service doesn't expose.

Can Fabric errors be caused by capacity limits?

Yes — Fabric capacities have concurrent operation limits. An undersized capacity causes failures during peak usage.

Does this error appear in Power BI Desktop?

No — Fabric items are cloud-native. Desktop can connect to Fabric semantic models but cannot trigger or observe Fabric-specific errors.

Source · learn.microsoft.com/en-us/fabric/data-factory/connector-troubleshoot-orc-format